Bug 105849
| Summary: | devel/gettext: [patch] Add MINIMAL option | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Alex Kozlov <spam> | ||||
| Component: | Individual Port(s) | Assignee: | Ade Lovett <ade> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Only Me | CC: | spam | ||||
| Priority: | Normal | ||||||
| Version: | Latest | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
|
Description
Alex Kozlov
2006-11-25 20:30:12 UTC
Responsible Changed From-To: freebsd-ports-bugs->ade Over to maintainer State Changed From-To: open->closed Whilst the "vast majority" of ports do not need the full gettext installation, the proposed "MINIMAL" patch is fundamentally flawed. 1. it will break those ports that *do* require extra bits and pieces 2. it saves minimal compile time, whilst heavily increased Makefile complexity. One could make a case for a gettext-minimal and gettext ports, however they'd conflict with each other, thus causing further pain. FreeBSD ports/packages are not like Linux distributions where there is generally a "runtime" and "development" package for each particular piece of software. On Mon, Nov 27, 2006 at 05:51:34AM +0000, Ade Lovett wrote: > Whilst the "vast majority" of ports do not need the full gettext installation, > the proposed "MINIMAL" patch is fundamentally flawed. > > 1. it will break those ports that *do* require extra bits and pieces This option for people, which know what they do. If I want to shoot myself in the foot, let me. > One could make a case for a gettext-minimal and gettext ports, however they'd > conflict with each other, thus causing further pain. Such was my primary plan. But it would demand modification USE_GETTEXT logic, that where more difficultly and less probably, what adding an option to port. p.s. If you do not want to work with this pr, please make it suspended. -- Adios On Nov 27, 2006, at 05:31 , Alex Kozlov wrote: > This option for people, which know what they do. > If I want to shoot myself in the foot, let me. You speak for yourself, which is fine. You do not speak for other people that select this option, and then hit my inbox when they run into a port that requires the full gettext installation. This is not fine. > Such was my primary plan. But it would demand modification > USE_GETTEXT logic, > that where more difficultly and less probably, what adding an > option to port. Ports don't work that way. You've doubled the complexity of the Makefile, for a grand savings of about 30 seconds of cpu time on my poor overworked Sempron 3100+, and further introduced a *large* minefield. > p.s. If you do not want to work with this pr, please make it > suspended. No. It is closed. I am the maintainer. End of story. Feel free to poke portmgr (as per policy) if you feel that you're being hard done by on this one. -aDe On Mon, Nov 27, 2006 at 05:42:37AM -0800, Ade Lovett wrote: > On Nov 27, 2006, at 05:31 , Alex Kozlov wrote: > >This option for people, which know what they do. > >If I want to shoot myself in the foot, let me. > You speak for yourself, which is fine. You do not speak > for other people that select this option, and then hit my > inbox when they run into a port that requires the full > gettext installation. This is not fine. Only if they select this option. Which can be provided with the proper warning or even excluded from OPTIONS. > >Such was my primary plan. But it would demand modification > >USE_GETTEXT logic, > >that where more difficultly and less probably, what adding an > >option to port. > Ports don't work that way. You've doubled the complexity of the > Makefile, for a grand savings of about 30 seconds of cpu time on my > poor overworked Sempron 3100+, and further introduced a *large* > minefield. You already introduced so much complications with EXAMPLES and HTMLMAN option, that superfluous 6 lines will change nothing. And this large minefield will exist only for those, who *wishes* to enter in him. > >p.s. If you do not want to work with this pr, please make it > >suspended. > No. It is closed. I am the maintainer. End of story. Feel free to > poke portmgr (as per policy) if you feel that you're being hard done > by on this one. Charmingly. -- Adios On Nov 27, 2006, at 07:27 , Alex Kozlov wrote: > Only if they select this option. Which can be provided with > the proper warning or even excluded from OPTIONS. Which still does not solve the issue of the non-trivial number of ports that assume that devel/gettext will provide a FULL installation, as opposed to the minimal installation you are proposing. > You already introduced so much complications with EXAMPLES and HTMLMAN > option, that superfluous 6 lines will change nothing. Wrong. The exclusions that occur from WITHOUT_EXAMPLES and WITHOUT_HTMLMAN do nothing to prevent a dependent port assuming that because devel/gettext is installed, *all* of the functionality of that port is available (beyond msgfmt and libintl). > And this large minefield will exist only for those, who *wishes* > to enter in him. Tell that to my inbox which will no doubt result in extra overhead (assuming a WITH_MINIMAL option is available), with random dependent ports suddenly failing, with no means of fixing it short of rebuilding devel/gettext without the MINIMAL flag set. There is plenty of other low-hanging fruit available for optimizing the ports infrastructure. Introducing a new option on devel/gettext, which is consumed by a *large* proportion of the remainder of the ports tree, without the capability to add in the rest of the functionality, is not it. Please try to think above and beyond the 30 seconds of extra compile time that this option saves, as opposed to the *large* amount of issues it will cause to the ports tree as a whole. -aDe |
