[libvirt PATCH] qemu: reject readonly attribute for virtiofs
This is not yet supported by virtiofsd. Fixes #23 Signed-off-by: Ján Tomko <jtomko@redhat.com> --- src/qemu/qemu_validate.c | 5 +++ .../vhost-user-fs-readonly.xml | 44 +++++++++++++++++++ tests/qemuxml2argvtest.c | 1 + 3 files changed, 50 insertions(+) create mode 100644 tests/qemuxml2argvdata/vhost-user-fs-readonly.xml diff --git a/src/qemu/qemu_validate.c b/src/qemu/qemu_validate.c index fde1892d42..ac22270104 100644 --- a/src/qemu/qemu_validate.c +++ b/src/qemu/qemu_validate.c @@ -3418,6 +3418,11 @@ qemuValidateDomainDeviceDefFS(virDomainFSDefPtr fs, return -1; case VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS: + if (fs->readonly) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("virtiofs does not yet supported read-only mode")); + return -1; + } if (!driver->privileged) { virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("virtiofs is not yet supported in session mode")); diff --git a/tests/qemuxml2argvdata/vhost-user-fs-readonly.xml b/tests/qemuxml2argvdata/vhost-user-fs-readonly.xml new file mode 100644 index 0000000000..003ed41eb3 --- /dev/null +++ b/tests/qemuxml2argvdata/vhost-user-fs-readonly.xml @@ -0,0 +1,44 @@ +<domain type='kvm'> + <name>guest</name> + <uuid>126f2720-6f8e-45ab-a886-ec9277079a67</uuid> + <memory unit='KiB'>14680064</memory> + <currentMemory unit='KiB'>14680064</currentMemory> + <memoryBacking> + <source type='file'/> + <access mode='shared'/> + </memoryBacking> + <vcpu placement='static'>2</vcpu> + <os> + <type arch='x86_64' machine='pc'>hvm</type> + <boot dev='hd'/> + </os> + <cpu mode='custom' match='exact' check='none'> + <model fallback='forbid'>qemu64</model> + <numa> + <cell id='0' cpus='0-1' memory='14680064' unit='KiB' memAccess='shared'/> + </numa> + </cpu> + <clock offset='utc'/> + <on_poweroff>destroy</on_poweroff> + <on_reboot>restart</on_reboot> + <on_crash>destroy</on_crash> + <devices> + <emulator>/usr/bin/qemu-system-x86_64</emulator> + <controller type='usb' index='0' model='none'/> + <controller type='pci' index='0' model='pci-root'/> + <filesystem type='mount' accessmode='passthrough'> + <driver type='virtiofs' queue='1024'/> + <binary path='/usr/libexec/virtiofsd' xattr='on'> + <cache mode='always'/> + <lock posix='off' flock='off'/> + </binary> + <source dir='/path'/> + <target dir='mount_tag'/> + <readonly/> + <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/> + </filesystem> + <input type='mouse' bus='ps2'/> + <input type='keyboard' bus='ps2'/> + <memballoon model='none'/> + </devices> +</domain> diff --git a/tests/qemuxml2argvtest.c b/tests/qemuxml2argvtest.c index 4ab664a846..8cbf3f8daf 100644 --- a/tests/qemuxml2argvtest.c +++ b/tests/qemuxml2argvtest.c @@ -3171,6 +3171,7 @@ mymain(void) DO_TEST_CAPS_LATEST("vhost-user-fs-fd-memory"); DO_TEST_CAPS_LATEST("vhost-user-fs-hugepages"); + DO_TEsT_CAPS_LATEST_PARSE_ERROR("vhost-user-fs-readonly"); DO_TEST("riscv64-virt", QEMU_CAPS_DEVICE_VIRTIO_MMIO); -- 2.25.4
On Wed, May 13, 2020 at 10:58:51AM +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Signed-off-by: Ján Tomko <jtomko@redhat.com> --- src/qemu/qemu_validate.c | 5 +++ .../vhost-user-fs-readonly.xml | 44 +++++++++++++++++++ tests/qemuxml2argvtest.c | 1 + 3 files changed, 50 insertions(+) create mode 100644 tests/qemuxml2argvdata/vhost-user-fs-readonly.xml
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here: https://gitlab.com/libvirt/libvirt/-/issues/23
case VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS: + if (fs->readonly) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("virtiofs does not yet supported read-only mode"));
s/supported/support/ Reviewed-by: Andrea Bolognani <abologna@redhat.com> -- Andrea Bolognani / Red Hat / Virtualization
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
case VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS: + if (fs->readonly) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("virtiofs does not yet supported read-only mode"));
s/supported/support/
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
-- Andrea Bolognani / Red Hat / Virtualization
Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
While I definitely don't like gitlab's URIs ...
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
... just mentioning #23 is useless for users looking at the git repo directly (not through the web-ui) or if we at some point decide to abandon gitlab. Even if gitlab can't handle the automation of closing the comment when the URI is mentioned I don't see why we should make it harder for users who want to stay away from the browser as much as possible.
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
While I definitely don't like gitlab's URIs ...
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
... just mentioning #23 is useless for users looking at the git repo directly (not through the web-ui) or if we at some point decide to abandon gitlab.
Even if gitlab can't handle the automation of closing the comment when the URI is mentioned I don't see why we should make it harder for users who want to stay away from the browser as much as possible.
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
While I definitely don't like gitlab's URIs ...
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
... just mentioning #23 is useless for users looking at the git repo directly (not through the web-ui) or if we at some point decide to abandon gitlab.
Even if gitlab can't handle the automation of closing the comment when the URI is mentioned I don't see why we should make it harder for users who want to stay away from the browser as much as possible.
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Even if they are flawed?! Just mentioning a random number in the commit message may be nice for people looking at the code via web-ui. That is nice but you can't do much there. But I and many other look at the code in the local checkout and 'git' doesn't randomly decide to expand the number to the gitlab uri and make it clickable. Now if I'd like to look at the issue after some time I'll e.g. when going through git-blame I'll have to open the browser, open gitlab, find the project and find the issue rather than just click a link. That's stupid. If gitlab can't deal with the link, it's frankly broken and we should not give in to a broken approach just because everybody else does.
On Wed, May 13, 2020 at 12:05:44PM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
While I definitely don't like gitlab's URIs ...
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
... just mentioning #23 is useless for users looking at the git repo directly (not through the web-ui) or if we at some point decide to abandon gitlab.
Even if gitlab can't handle the automation of closing the comment when the URI is mentioned I don't see why we should make it harder for users who want to stay away from the browser as much as possible.
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Even if they are flawed?! Just mentioning a random number in the commit message may be nice for people looking at the code via web-ui. That is nice but you can't do much there. But I and many other look at the code in the local checkout and 'git' doesn't randomly decide to expand the number to the gitlab uri and make it clickable.
Now if I'd like to look at the issue after some time I'll e.g. when going through git-blame I'll have to open the browser, open gitlab, find the project and find the issue rather than just click a link. That's stupid. If gitlab can't deal with the link, it's frankly broken and we should not give in to a broken approach just because everybody else does.
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 12:05:44PM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
[...]
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Even if they are flawed?! Just mentioning a random number in the commit message may be nice for people looking at the code via web-ui. That is nice but you can't do much there. But I and many other look at the code in the local checkout and 'git' doesn't randomly decide to expand the number to the gitlab uri and make it clickable.
Now if I'd like to look at the issue after some time I'll e.g. when going through git-blame I'll have to open the browser, open gitlab, find the project and find the issue rather than just click a link. That's stupid. If gitlab can't deal with the link, it's frankly broken and we should not give in to a broken approach just because everybody else does.
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above: 1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring. Just observe the following behaviours on the commit that was just pushed and Jano helpfully for this demonstration included both versions: 1) gitlab: libvirt/libvirt https://gitlab.com/libvirt/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3... Here I'd expect a clickable link but there isn't one. I have no idea why. You can see that gitlab doesn't recognize it. I'm not sure whether the format Jano chose is bad, but this already enforces some kind of form we need to conform to. 2) gitlab: pipo.sk/libvirt https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3... Unsurprisingly it doesn't work here, but the full link is shortened and still points to the correct issue. 3) github: libvirt/libvirt https://github.com/libvirt/libvirt/commit/9c58b6eb0090b978f44299b22ff4f8d3be... As expected #23 points to nothing, but the full link still points to the issue even cross-hosting. 4) github: pipo/libvirt (plus I've created 23 test issues in my fork) https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3... Now this is fun. '#23' points to the local issue #23 in my fork while the full link still points to the issue. The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
On Wed, May 13, 2020 at 13:06:58 +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 12:05:44PM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
[...]
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Even if they are flawed?! Just mentioning a random number in the commit message may be nice for people looking at the code via web-ui. That is nice but you can't do much there. But I and many other look at the code in the local checkout and 'git' doesn't randomly decide to expand the number to the gitlab uri and make it clickable.
Now if I'd like to look at the issue after some time I'll e.g. when going through git-blame I'll have to open the browser, open gitlab, find the project and find the issue rather than just click a link. That's stupid. If gitlab can't deal with the link, it's frankly broken and we should not give in to a broken approach just because everybody else does.
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above:
1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues
Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring.
Just observe the following behaviours on the commit that was just pushed and Jano helpfully for this demonstration included both versions:
1) gitlab: libvirt/libvirt
https://gitlab.com/libvirt/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Here I'd expect a clickable link but there isn't one. I have no idea why.
You can see that gitlab doesn't recognize it. I'm not sure whether the format Jano chose is bad, but this already enforces some kind of form we need to conform to.
2) gitlab: pipo.sk/libvirt
https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Unsurprisingly it doesn't work here, but the full link is shortened and still points to the correct issue.
3) github: libvirt/libvirt
https://github.com/libvirt/libvirt/commit/9c58b6eb0090b978f44299b22ff4f8d3be...
As expected #23 points to nothing, but the full link still points to the issue even cross-hosting.
4) github: pipo/libvirt (plus I've created 23 test issues in my fork)
https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Oops, messed up the most interesting one: https://github.com/pipo/libvirt/commit/9c58b6eb0090b978f44299b22ff4f8d3be3a2...
Now this is fun. '#23' points to the local issue #23 in my fork while the full link still points to the issue.
The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
On Wed, 2020-05-13 at 13:06 +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above:
1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues
Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring.
[...]
The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
I completely agree, #nnn is too ambiguous to be useful. It seems that GitLab happily accepts full URLs instead of shortened identifiers: https://docs.gitlab.com/ee/user/project/issues/managing_issues.html#default-... We can have a simple prebuild check, similar to the one we already use for DCO checking, which catches uses of Fixes #nnn and similar and tells contributors to use a full URL instead. -- Andrea Bolognani / Red Hat / Virtualization
On a Wednesday in 2020, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 13:06 +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above:
1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues
Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring.
[...]
The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
I completely agree, #nnn is too ambiguous to be useful.
It seems that GitLab happily accepts full URLs instead of shortened identifiers:
https://docs.gitlab.com/ee/user/project/issues/managing_issues.html#default-...
We can have a simple prebuild check, similar to the one we already use for DCO checking, which catches uses of
Fixes #nnn
and similar and tells contributors to use a full URL instead.
For DCO, the idea is that you specifically need the user to act on it. If we switch to merge requests, this coudl be handled by the same bot that applies the R-b tags, right? :) Jano
On Wed, May 13, 2020 at 01:19:35PM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 13:06 +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above:
1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues
Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring.
[...]
The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
I completely agree, #nnn is too ambiguous to be useful.
The widespread usage by any other project using GitLab/GitHub proves otherwise and libvirt isn't special in this regard. We're all smart enough to understand this.
We can have a simple prebuild check, similar to the one we already use for DCO checking, which catches uses of
Fixes #nnn
and similar and tells contributors to use a full URL instead.
This is needless extra work for contributors. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, May 13, 2020 at 12:30:48 +0100, Daniel P. Berrangé wrote:
On Wed, May 13, 2020 at 01:19:35PM +0200, Andrea Bolognani wrote: ...
I completely agree, #nnn is too ambiguous to be useful.
The widespread usage by any other project using GitLab/GitHub proves otherwise and libvirt isn't special in this regard.
This just means that either everyone uses the reference only as a simple way of closing the associated issue or they use the web UI to look at commits.
We're all smart enough to understand this.
It's not really about understanding how to get to the issue. It's about the effort needed to find the issue. I guess for most commits with the short reference looking at the original issue would not be worth the effort of finding it. Which is fine, we can safe time by not looking at unimportant links.
We can have a simple prebuild check, similar to the one we already use for DCO checking, which catches uses of
Fixes #nnn
and similar and tells contributors to use a full URL instead.
This is needless extra work for contributors.
I guess I'm a bit strange, but to me pasting a complete link to an issue is (if I even know such issue exists, of course) less work than selecting just the number and prefixing it with '#'. Either way I think fighting about such insignificant point is not the best use of our time. Let's save it for more important battles we're going to face, such as the switch to merge requests. Jirka
On Wed, 2020-05-13 at 15:00 +0200, Jiri Denemark wrote:
On Wed, May 13, 2020 at 12:30:48 +0100, Daniel P. Berrangé wrote:
On Wed, May 13, 2020 at 01:19:35PM +0200, Andrea Bolognani wrote:
I completely agree, #nnn is too ambiguous to be useful.
The widespread usage by any other project using GitLab/GitHub proves otherwise and libvirt isn't special in this regard.
This just means that either everyone uses the reference only as a simple way of closing the associated issue or they use the web UI to look at commits.
Let's just hope they don't use the "wrong" Web UI, for example by looking at the GitHub mirror, because otherwise they might be sent to a completely wrong issue, as Peter just demonstrated.
We can have a simple prebuild check, similar to the one we already use for DCO checking, which catches uses of
Fixes #nnn
and similar and tells contributors to use a full URL instead.
This is needless extra work for contributors.
I guess I'm a bit strange, but to me pasting a complete link to an issue is (if I even know such issue exists, of course) less work than selecting just the number and prefixing it with '#'.
If that's the case, then at least know that you're not the only one who's strange around here :) -- Andrea Bolognani / Red Hat / Virtualization
On Wed, May 13, 2020 at 01:06:58PM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 12:05:44PM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
[...]
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Even if they are flawed?! Just mentioning a random number in the commit message may be nice for people looking at the code via web-ui. That is nice but you can't do much there. But I and many other look at the code in the local checkout and 'git' doesn't randomly decide to expand the number to the gitlab uri and make it clickable.
Now if I'd like to look at the issue after some time I'll e.g. when going through git-blame I'll have to open the browser, open gitlab, find the project and find the issue rather than just click a link. That's stupid. If gitlab can't deal with the link, it's frankly broken and we should not give in to a broken approach just because everybody else does.
It isn't about giving in. Again the point is to not needlessly create special rules for contributing to libvirt, because every special rule we add is another thing for contributors to stumble over. Some rules are worth it because they have meaningful benefits such as the use of Signed-off-by/DCO. The mentioning of full URLs instead of the normal issue reference syntax does not have a meaningful benefit that justifies a libvirt special rule for contributions.
I gave an examples of two specific meaningful benefit above:
1) it provides a clickable link without second guessing where to go for command line users 2) provides stable reference to the hosting of issues
Note that for example github uses exactly the same format for referencing issues. That means that it's unclear what we are referring.
Just observe the following behaviours on the commit that was just pushed and Jano helpfully for this demonstration included both versions:
1) gitlab: libvirt/libvirt
https://gitlab.com/libvirt/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Here I'd expect a clickable link but there isn't one. I have no idea why.
You can see that gitlab doesn't recognize it. I'm not sure whether the format Jano chose is bad, but this already enforces some kind of form we need to conform to.
2) gitlab: pipo.sk/libvirt
https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Unsurprisingly it doesn't work here, but the full link is shortened and still points to the correct issue.
3) github: libvirt/libvirt
https://github.com/libvirt/libvirt/commit/9c58b6eb0090b978f44299b22ff4f8d3be...
As expected #23 points to nothing, but the full link still points to the issue even cross-hosting.
4) github: pipo/libvirt (plus I've created 23 test issues in my fork)
https://gitlab.com/pipo.sk/libvirt/-/commit/9c58b6eb0090b978f44299b22ff4f8d3...
Now this is fun. '#23' points to the local issue #23 in my fork while the full link still points to the issue.
The shortened issue names are ambiguous and the hosting has no way in figuring out where to point to. Providing full URL is not something which should be described as "no meaningful benefit" but it actively disambiguates the links regardless of where it's hosted or refered from.
Ok, I'll withdraw my objection to use of full URLs. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On a Wednesday in 2020, Daniel P. Berrangé wrote:
On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
This is not yet supported by virtiofsd.
Fixes #23
Please include the full URL here:
While I definitely don't like gitlab's URIs ...
Using the '#23' format is the recommended way. GitLab UI will turn it into a hyperlink, and automatically close the mentioned issue.
... just mentioning #23 is useless for users looking at the git repo directly (not through the web-ui) or if we at some point decide to abandon gitlab.
Even if gitlab can't handle the automation of closing the comment when the URI is mentioned I don't see why we should make it harder for users who want to stay away from the browser as much as possible.
The point is for libvirt to follow normal practice from GitLab, so that contributors don't have to know about libvirt specific rules for contributing to the project. Telling users to change the normal issue syntax into a URL whenever they send a patch is not something we should be doing.
Okay, I'll roll a die and push it one of with: 1) no issue reference 2) #23 3) https://gitlab.com/libvirt/libvirt/-/issues/23 4) #23 and https://gitlab.com/libvirt/libvirt/-/issues/23 5) https://www.youtube.com/watch?v=dQw4w9WgXcQ 6) https://gitlab.com/libvirt/libvirt/-/issues/23 and #23 ... Jano
Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
participants (6)
-
Andrea Bolognani -
Daniel P. Berrangé -
Jiri Denemark -
Ján Tomko -
Peter Krempa -
Pino Toscano